To achieve carbon neutrality by 2050, grasping actual energy consumption at the urban level and establishing efficient management methodologies have become urgent global imperatives. However, in Japan, collecting granular data such as insulation performance and equipment specifications for individual buildings is challenging, creating a significant barrier to wide-area energy assessments. To address this issue, this study developed a methodology to automatically generate EnergyPlus simulation models from building footprints and heights by utilizing open data from the 3D city model "Project PLATEAU". This paper reports the verification of the system's applicability to a residential district in Sapporo, a cold region in Japan. First, accuracy verification using measured data from two existing houses confirmed that heating load predictions from the automatically generated models agreed well with measured values. Subsequently, the method was extended to 3,420 residential buildings in the target area. Multifaceted statistical analyses, including multiple linear regression and Random Forest, quantitatively demonstrated that physical variables such as total envelope area and ventilation rates are the dominant factors governing heating loads. Furthermore, the model exhibited appropriate sensitivity to solar heat gain, aligning with actual thermal phenomena. These results demonstrate that the simulation model functions effectively at a scale of over 3,000 buildings without anomalous outliers. This methodology serves as a robust support tool for strengthening urban energy resilience and formulating evidence-based decarbonization roadmaps.
Osawa et al. (Tue,) studied this question.